This study showed no association between s’, global strain, LVEF, TAPSE, or E/e’ acquired after reaching the target temperature and 180 days neurological outcome. No echocardiography parameter showed consistent differences across all scan time points. Higher e’ and ΔIVC 48 h after initiation of TTM were associated with good neurological outcome at 180 days follow-up amongst comatose survivors of OHCA
TTE evaluation of cardiac function
In our study, e’ and ΔIVC at 48 h were the only TTE parameters associated with neurological outcome. The lower IVC variability found during hypothermia may reflect an elevated preload caused by diastolic dysfunction during hypothermia (24). At 48 h, the TTM48 group was still hypothermic and the TTM24 group was normothermic adding to random variation at this point. At 72 h, more patients in the poor neurological outcome group were intubated and received positive pressure ventilation. Positive pressure ventilation is known to affect diastolic echocardiography parameters (25) and IVC dynamics (26). In our study the duration of positive pressure ventilation was longer in the TTM48 group due to the prolonged cooling period compared with the TTM24 group. The prolonged positive pressure ventilation in the TTM48 group could be the reason for the lower e’ and ΔIVC at 48 h. However, these individual comparisons were not adjusted for the number of analyses for each echocardiography variable. The overall analyses for all echocardiography variables did not show any significant differences between groups and, hence, the significant differences at any individual time points are probably type 1 errors. A recent study found that low e’ was associated with in-hospital death following OHCA (18). Significantly more patients with poor outcome were, however, intubated and thus the results may be confounded by the effect of positive pressure ventilation on diastolic function (18).
We previously showed a beneficial effect of TTM48 on s’ compared with TTM24; however, this effect was not seen across other echocardiography parameters (13). Our study did not show an association between echocardiography measures of left ventricular systolic function and outcome. We included s’ and GLS in our evaluation of left ventricular systolic function since these markers are less preload-dependent and more sensitive to changes in left ventricular function compared with LVEF, especially following ischemia (27–31). In general, GLS may overestimate left ventricular function during ischaemia due to post-systolic shortening, adding to random variation (30). However, as we defined the end of the systole as the closure of the aortic valve, post-systolic shortening should not have affected the GLS values included in this study. Heart rate increased during rewarming and may have affected speckle tracking and thus the GLS measurements (28).
Previously, we showed that s’ improved in patients treated with TTM48 (13) although TTM48 did not improve neurological outcome compared with TTM24 (12). s’ was not associated with neurological outcome in our study, despite being highly reproducible and sensitive to changes in left ventricular systolic function and insensitive to poor image quality often encountered in intensive care patients (29, 32). However, s’ is angle-dependent and measurements may be affected by respiration (25, 29).
Timing of TTE and clinical implications
Our study evaluates the cooling and peri-rewarming phase during post-cardiac arrest TTM. PCAMD is present within the first hours following cardiac arrest and slowly recovers in the following days and months (4, 17). LVEF, E/e’ and right ventricular diameter assessed following cardiac arrest have been shown to be associated with mortality and neurological outcome (14–17, 19, 33–35), though results are inconsistent (18, 36–38). In studies showing an association between LVEF, E/e’, and right ventricular function TTE was performed within the first 24 h of admission to hospital (14–17, 19, 33–35). In this period, the immediate effects of acute ischaemia of the heart and affected ventricular function are more pronounced (4, 17). Patients in previous studies were heterogeneous and not comparable to our population with regard to lower rates of bystander CPR, longer times to ROSC, fewer patients had a primary shockable rhythm and survival rates were lower (14–17, 19, 33–35). Patients in our study were immediately evaluated with coronary angiography and significant coronary lesions were treated. Thus, previous study populations have included more affected patients in the acute phase following cardiac ischaemia. Existing studies suggest that assessment of acute ischemia on left ventricular function has a greater potential of predicting neurological outcome compared with later scan times. However, the present study suggests that early serial TTE evaluation of cardiac function is not effective in prognosticating neurological outcome in the cooling and peri-rewarming phases following OHCA.
In our study, patients with good neurological outcome more often had a primary shockable rhythm, shorter times to ROSC, lower arrival lactate, and lower SAPSII scores (Table 1) indicating that prognostication of outcome following OHCA is a multimodal process including several clinical parameters (4). Our data suggest that assessment of PCAMD on early serial TTEs following OHCA does not add important clinical value in prognostication of neurological outcome following cardiac arrest.
Our study has certain strengths as opposed to previous studies. We present results based on standardised early serial TTE scan times in a well-defined population of comatose OHCA patients. Standardised scan times were utilised to ensure comparability between data. Thus, our design enabled assessment of changes in TTE parameters over time and not only at single time point estimates. Few and experienced doctors performed all echocardiographies. One person blinded to patient treatment and outcome did the offline analyses, and inter- and intra-observer variability have previously been shown to be low (13). Nonetheless, the process of patient inclusion in this study may have been subject to selection bias. Written consent was obtained from a legal next of kin following admission, which may have delayed randomisation slightly and, hence, the most ill patients may have been lost before study enrolment. As cardiac death often occurs early, this may have impacted on our results. In contrast to previous studies (18, 19), we excluded patients in prolonged cardiogenic shock, thus our selection process has probably favoured a population less ill compared to previous studies. Patients with good outcome more often had a primary shockable rhythm and were less likely to be diagnosed with diabetes (Table 1). Due to the number of patients in our analysis, we decided to adjust for a maximum of three variables (primary rhythm, time to ROSC and age). The rate of bystander CPR of 88% in our population was higher compared with previous studies (14, 18, 19, 39), but represent similar rates reported in a Danish population in the TTM trial (40). Only 34 patients had a poor neurological outcome limiting multivariate analysis and mixed model analysis in each group. Our analysis was a post-hoc analysis and power calculations have thus not been performed; type 2 error was a risk.
The scan time points in our study were standardised during cooling and rewarming. In an effort to limit missing data, three scan time points were chosen and thus, the TTM24 group had two scans during rewarming while the TTM48 group only had one. Comparing different rewarming periods may have added random variation to our results.